Commit 269dcc1c2ec25864308ee03a3fa26ea819d9f5d0
1 parent
0ddea0ead2
Exists in
master
and in
4 other branches
rcu: Add tracing data to support queueing models
The current tracing data is not sufficient to deduce the average time that a callback spends waiting for a grace period to end. Add three per-CPU counters recording the number of callbacks invoked (ci), the number of callbacks orphaned (co), and the number of callbacks adopted (ca). Given the existing callback queue length (ql), the average wait time in absence of CPU hotplug operations is ql/ci. The units of wait time will be in terms of the duration over which ci was measured. In the presence of CPU hotplug operations, there is room for argument, but ql/(ci-co+ca) won't steer you too far wrong. Also fixes a typo called out by Lucas De Marchi <lucas.de.marchi@gmail.com>. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Showing 4 changed files with 25 additions and 4 deletions Side-by-side Diff
Documentation/RCU/trace.txt
| ... | ... | @@ -125,6 +125,17 @@ |
| 125 | 125 | of RCU callbacks is ready to invoke, then the remainder will |
| 126 | 126 | be deferred. |
| 127 | 127 | |
| 128 | +o "ci" is the number of RCU callbacks that have been invoked for | |
| 129 | + this CPU. Note that ci+ql is the number of callbacks that have | |
| 130 | + been registered in absence of CPU-hotplug activity. | |
| 131 | + | |
| 132 | +o "co" is the number of RCU callbacks that have been orphaned due to | |
| 133 | + this CPU going offline. | |
| 134 | + | |
| 135 | +o "ca" is the number of RCU callbacks that have been adopted due to | |
| 136 | + other CPUs going offline. Note that ci+co-ca+ql is the number of | |
| 137 | + RCU callbacks registered on this CPU. | |
| 138 | + | |
| 128 | 139 | There is also an rcu/rcudata.csv file with the same information in |
| 129 | 140 | comma-separated-variable spreadsheet format. |
| 130 | 141 | |
| ... | ... | @@ -180,7 +191,7 @@ |
| 180 | 191 | |
| 181 | 192 | o "jfq" is the number of jiffies remaining for this grace period |
| 182 | 193 | before force_quiescent_state() is invoked to help push things |
| 183 | - along. Note that CPUs in dyntick-idle mode thoughout the grace | |
| 194 | + along. Note that CPUs in dyntick-idle mode throughout the grace | |
| 184 | 195 | period will not report on their own, but rather must be check by |
| 185 | 196 | some other CPU via force_quiescent_state(). |
| 186 | 197 |
kernel/rcutree.c
| ... | ... | @@ -1004,6 +1004,7 @@ |
| 1004 | 1004 | for (i = 0; i < RCU_NEXT_SIZE; i++) |
| 1005 | 1005 | rdp->nxttail[i] = &rdp->nxtlist; |
| 1006 | 1006 | rsp->orphan_qlen += rdp->qlen; |
| 1007 | + rdp->n_cbs_orphaned += rdp->qlen; | |
| 1007 | 1008 | rdp->qlen = 0; |
| 1008 | 1009 | raw_spin_unlock(&rsp->onofflock); /* irqs remain disabled. */ |
| 1009 | 1010 | } |
| ... | ... | @@ -1025,6 +1026,7 @@ |
| 1025 | 1026 | *rdp->nxttail[RCU_NEXT_TAIL] = rsp->orphan_cbs_list; |
| 1026 | 1027 | rdp->nxttail[RCU_NEXT_TAIL] = rsp->orphan_cbs_tail; |
| 1027 | 1028 | rdp->qlen += rsp->orphan_qlen; |
| 1029 | + rdp->n_cbs_adopted += rsp->orphan_qlen; | |
| 1028 | 1030 | rsp->orphan_cbs_list = NULL; |
| 1029 | 1031 | rsp->orphan_cbs_tail = &rsp->orphan_cbs_list; |
| 1030 | 1032 | rsp->orphan_qlen = 0; |
| ... | ... | @@ -1156,6 +1158,7 @@ |
| 1156 | 1158 | |
| 1157 | 1159 | /* Update count, and requeue any remaining callbacks. */ |
| 1158 | 1160 | rdp->qlen -= count; |
| 1161 | + rdp->n_cbs_invoked += count; | |
| 1159 | 1162 | if (list != NULL) { |
| 1160 | 1163 | *tail = rdp->nxtlist; |
| 1161 | 1164 | rdp->nxtlist = list; |
kernel/rcutree.h
| ... | ... | @@ -202,6 +202,9 @@ |
| 202 | 202 | long qlen; /* # of queued callbacks */ |
| 203 | 203 | long qlen_last_fqs_check; |
| 204 | 204 | /* qlen at last check for QS forcing */ |
| 205 | + unsigned long n_cbs_invoked; /* count of RCU cbs invoked. */ | |
| 206 | + unsigned long n_cbs_orphaned; /* RCU cbs sent to orphanage. */ | |
| 207 | + unsigned long n_cbs_adopted; /* RCU cbs adopted from orphanage. */ | |
| 205 | 208 | unsigned long n_force_qs_snap; |
| 206 | 209 | /* did other CPU force QS recently? */ |
| 207 | 210 | long blimit; /* Upper limit on a processed batch */ |
kernel/rcutree_trace.c
| ... | ... | @@ -64,7 +64,9 @@ |
| 64 | 64 | rdp->dynticks_fqs); |
| 65 | 65 | #endif /* #ifdef CONFIG_NO_HZ */ |
| 66 | 66 | seq_printf(m, " of=%lu ri=%lu", rdp->offline_fqs, rdp->resched_ipi); |
| 67 | - seq_printf(m, " ql=%ld b=%ld\n", rdp->qlen, rdp->blimit); | |
| 67 | + seq_printf(m, " ql=%ld b=%ld", rdp->qlen, rdp->blimit); | |
| 68 | + seq_printf(m, " ci=%lu co=%lu ca=%lu\n", | |
| 69 | + rdp->n_cbs_invoked, rdp->n_cbs_orphaned, rdp->n_cbs_adopted); | |
| 68 | 70 | } |
| 69 | 71 | |
| 70 | 72 | #define PRINT_RCU_DATA(name, func, m) \ |
| ... | ... | @@ -119,7 +121,9 @@ |
| 119 | 121 | rdp->dynticks_fqs); |
| 120 | 122 | #endif /* #ifdef CONFIG_NO_HZ */ |
| 121 | 123 | seq_printf(m, ",%lu,%lu", rdp->offline_fqs, rdp->resched_ipi); |
| 122 | - seq_printf(m, ",%ld,%ld\n", rdp->qlen, rdp->blimit); | |
| 124 | + seq_printf(m, ",%ld,%ld", rdp->qlen, rdp->blimit); | |
| 125 | + seq_printf(m, ",%lu,%lu,%lu\n", | |
| 126 | + rdp->n_cbs_invoked, rdp->n_cbs_orphaned, rdp->n_cbs_adopted); | |
| 123 | 127 | } |
| 124 | 128 | |
| 125 | 129 | static int show_rcudata_csv(struct seq_file *m, void *unused) |
| ... | ... | @@ -128,7 +132,7 @@ |
| 128 | 132 | #ifdef CONFIG_NO_HZ |
| 129 | 133 | seq_puts(m, "\"dt\",\"dt nesting\",\"dn\",\"df\","); |
| 130 | 134 | #endif /* #ifdef CONFIG_NO_HZ */ |
| 131 | - seq_puts(m, "\"of\",\"ri\",\"ql\",\"b\"\n"); | |
| 135 | + seq_puts(m, "\"of\",\"ri\",\"ql\",\"b\",\"ci\",\"co\",\"ca\"\n"); | |
| 132 | 136 | #ifdef CONFIG_TREE_PREEMPT_RCU |
| 133 | 137 | seq_puts(m, "\"rcu_preempt:\"\n"); |
| 134 | 138 | PRINT_RCU_DATA(rcu_preempt_data, print_one_rcu_data_csv, m); |